Addictive drugs are best known for their rewarding properties, but also influence behavioral responses to aversive stimuli;for example, cocaine can initially suppress aversive processes, while subsequently inducing delayed aversive crashes. We hypothesized these biphasic effects involve the lateral habenula (LHb) and its major target, the rostromedial tegmentum (RMTg), two structures that process aversive information and are reciprocally linked with dopamine neurons. Biphasic responses of LHb neurons by cocaine Intravenous (iv) cocaine at a dose of 0.75mg/kg has previously been shown to be rewarding during a period 0-5 minutes after injections, and aversive during a period 15-20 minutes after injections. To determine whether LHb responses to cocaine parallel these motivational shifts, we recorded 20 LHb units in 8 awake behaving rats receiving either iv saline or cocaine. During the period 0-5 minutes after injections, 80% of these neurons were significantly inhibited by cocaine, a proportion significantly higher than chance (p = 8 x 10-16, binomial distribution), while only one neuron was excited in this period, a proportion not different from chance (p = 0.66, binomial distribution). Firing rates of the inhibited neurons were reduced to 57 4% of baseline during the period 0-5 minutes post-injection, and remained significantly inhibited 5-10 minutes post-injection. In contrast, during a period 15-30 minutes post-injection, 5 of the 20 neurons were significantly excited by cocaine, a proportion again higher than chance (p = 0.003, binomial distribution), while only one neuron was significantly inhibited, a proportion no different from chance (p = 0.66). A 6th neuron showed a significant excitation in the 10-15 minute post-injection bin, just before the 15-30 minute window when the other units were activated, suggesting some heterogeneity in the exact time of excitation. Hence, around 30% (6/20) of recorded LHb neurons showed some form of delayed excitations to cocaine. In most neurons, excitations lasted up to 30 minutes post-injection, when cocaine levels are predicted to have fallen considerably. Hence, the delayed excitation is not likely a direct pharmacological effect of cocaine, but could be a compensatory response. Saline iv injections produced no significant effects in either time window on either individual or average firing rates (p >0.05, all comparisons). Cocaine-responsive LHb neurons were distributed throughout the nucleus and tended to reside near areas of dense tyrosine hydroxylase innervation in the LHb. Average firing rates of nearby thalamic units (n = 11) were not affected by cocaine in either the earlier (p = 0.78) or later time windows (p = 0.4). Food predictive cues To determine whether LHb neuron responses to cocaine would generalize more broadly to other salient stimuli, as suggested by primate studies 23,35, we examined the responses of some LHb neurons to both cocaine infusions and to food-predictive auditory cues. Of 16 units for which responses to both stimuli were available, 7 units showed significant responses to both stimuli. In all 7 cases, the direction of modulation was inhibitory to both, consistent with similar encodings of drug and food rewards. One of these neurons was also excited by an auditory cue predicting non-reward. These patterns are consistent with encoding general information about affective valence, rather than the specific sensory properties of particular stimuli. In summary, we showed in awake behaving rats that some LHb neurons show biphasic responses to intravenous cocaine that temporally coincide with a shift from rewarding to aversive effects of the drug.